prepare datasets only flag
This commit is contained in:
@@ -306,6 +306,7 @@ def setup_trainer(cfg, train_dataset, eval_dataset, model, tokenizer):
|
|||||||
|
|
||||||
def train(
|
def train(
|
||||||
config: Path = Path("configs/"),
|
config: Path = Path("configs/"),
|
||||||
|
prepare_ds_only: bool = False,
|
||||||
**kwargs,
|
**kwargs,
|
||||||
):
|
):
|
||||||
if Path(config).is_dir():
|
if Path(config).is_dir():
|
||||||
@@ -396,6 +397,10 @@ def train(
|
|||||||
else:
|
else:
|
||||||
dataset.save_to_disk(DEFAULT_DATASET_PREPARED_PATH)
|
dataset.save_to_disk(DEFAULT_DATASET_PREPARED_PATH)
|
||||||
|
|
||||||
|
if prepare_ds_only:
|
||||||
|
logger.info("Finished preparing dataset. Exiting...")
|
||||||
|
return
|
||||||
|
|
||||||
train_dataset = dataset["train"]
|
train_dataset = dataset["train"]
|
||||||
eval_dataset = dataset["test"]
|
eval_dataset = dataset["test"]
|
||||||
|
|
||||||
|
|||||||
Reference in New Issue
Block a user